Abstract
This paper describes an outdoor localization method by matching skyline. On a surface of the moon, we cannot use either the GNSS (Global Navigation Satellite System), information of artificial landmark or map matching. However a lot of planetary probe's missions need absolute coordinates. Therefore we propose a method that uses DEM (Digital Elevation Model) and panoramic photo image, and compares skylines made by them. Extracting skylines from image uses k-means clustering method with color information. A skyline made by panoramic photo image is determined uniquely by 4 parameters. However the proposed method fixes 2 parameters and uses a rough calculation, assuming that the place to get the photo image is horizontality. As the result of experiments at Oshima Island, it was confirmed that it is possible to estimate location in error by 65 m or 19 m by using the proposed method. One of prospective causes of an error is the assumption of horizontality in the place getting photo images, so we expect to improve accuracy of estimating location if this method is used with attitude sensor.
